During the last glacial/interglacial transition the Earth's climate underwent abrupt changes around 14.6 kyr ago. Temperature proxies from ice cores revealed the onset of the Bølling/Allerød (B/A) warm period in the north and the start of the Antarctic Cold Reversal in the south. Furthermore, the B/A was accompanied by a rapid sea level rise of about 20 m during meltwater pulse (MWP) 1A, whose exact timing is a matter of current debate. In-situ measured CO<sub>2</sub> in the EPICA Dome C (EDC) ice core also revealed a remarkable jump of 10 ± 1 ppmv in 230 yr at the same time. Allowing for the modelled age distribution of CO<sub>2</sub> in firn, we show that atmospheric CO<sub>2</sub> could have jumped by 20–35 ppmv in less than 200 yr, which is a factor of 2–3.5 greater than the CO<sub>2</sub> signal recorded in-situ in EDC. This rate of change in atmospheric CO<sub>2</sub> corresponds to 29–50% of the anthropogenic signal during the last 50 yr and is connected with a radiative forcing of 0.59–0.75 W m<sup>−2</sup>. Using a model-based airborne fraction of 0.17 of atmospheric CO<sub>2</sub>, we infer that 125 Pg of carbon need to be released into the atmosphere to produce such a peak. If the abrupt rise in CO<sub>2</sub> at the onset of the B/A is unique with respect to other Dansgaard/Oeschger (D/O) events of the last 60 kyr (which seems plausible if not unequivocal based on current observations), then the mechanism responsible for it may also have been unique. Available δ<sup>13</sup>CO<sub>2</sub> data are neutral, whether the source of the carbon is of marine or terrestrial origin. We therefore hypothesise that most of the carbon might have been activated as a consequence of continental shelf flooding during MWP-1A. This potential impact of rapid sea level rise on atmospheric CO<sub>2</sub> might define the point of no return during the last deglaciation.